Spring 2023

April 2005

MassGIS 2023 Aerial Imagery Dataset

  Dataset Description

This dataset contains high-resolution "leaf off" aerial photographs covering the entire state of Massachusetts, captured in spring 2023. The imagery has been converted from specialized formats into standard formats suitable for AI and machine learning applications.

  Overview

The MassGIS 2023 Aerial Imagery Dataset provides comprehensive aerial photography coverage of Massachusetts, captured during leaf-off… See the full description on the dataset page: https://huggingface.co/datasets/evijit/massgis_2023_aerial_imagery.

Spring 2023 "true color" aerial imagery for Massachusetts, published as a tile layer at ArcGIS Online by MassGIS.This layer is based on 8-bit, 15 cm resolution JPEG 2000 versions of GeoTiff imagery, using the Red-Green-Blue bands.Funding for this imagery was provided by MassDOT, the State 911 Department, and the Executive Office of Technology Services and Security (EOTSS).This digital orthoimagery was created to provide easily accessible geospatial data which are readily available to enhance the capability of federal, state, and local emergency responders, as well as to plan for homeland security efforts. These data also support The National Map.These images can serve a variety of purposes, from general planning to field reference for spatial analysis, to a tool for data development and revision of vector maps. The imagery can also serve as a reference layer or basemap for myriad applications inside geographic information system (GIS) software and web-based maps.More details...

The U.S. Geological Survey (USGS) Remote Sensing Coastal Change (RSCC) project collects aerial imagery along coastal swaths with optimized endlap/sidelap and precise position information to create high-resolution orthomosaics, three-dimensional (3D) point clouds, and digital elevation/surface models (DEMs/DSMs) using Structure-from-Motion (SfM) photogrammetry methods. These products are valuable for measuring topographic and landscape change, and for understanding coastal vulnerability and response to disturbance events. A nadir (vertical) aerial imagery survey was conducted on Cape Cod, Massachusetts in two sections November 4th, 2023, to document the state of the coast. The first section covers the coast on the Cape Cod Bay from Plymouth to Orleans and the second section covers the coast of the Outer Cape facing the Atlantic Ocean from Provincetown to Monomoy Island. The observations along the coastline cover an approximately 140-kilometer-long by 300 to 700-meter-wide swath of coastline and encompass both highly developed towns as well as natural undeveloped areas, including the federal lands of Cape Cod National Seashore. Low altitude (300 meters above ground level) digital aerial imagery were acquired with a manned, fixed-wing aircraft using the "Precision Airborne Camera (PAC)" System (version 2). The PAC system is operated by C.W. Wright and consists of a mounted fixed-lens digital camera, along with a custom integrated survey-grade Global Navigation Satellite System (GNSS) receiver. Data were collected in shore-parallel lines, flying at approximately 50 meters per second (m/s) and capturing true color imagery at 1 hertz (Hz), resulting in image footprints with approximately 75-80% endlap, 60-70% sidelap, and a 5.3-centimeter (cm) ground sample distance (GSD). The precise time of each image capture (flash event) was recorded, and the corresponding aircraft position was computed during post-processing of the GNSS data. Precise image positions can then be determined by accounting for the lever arm offsets between the aircraft GNSS antenna and the camera lens, which are provided in the PAC System metadata (Kranenburg and others, 2023, https://cmgds.marine.usgs.gov/data-services/rscc/PrecisionAirborneCameraSystem). Position data, provided as latitude/longitude/ellipsoid height, is referenced to the North American Datum of 1983 National Spatial Reference System 2011 (NAD83(2011)) coordinate system.

Context

Building Segmentation from Aerial Imagery is a challenging task. Obstruction from nearby trees, shadows of adjacent buildings, varying texture and color of rooftops, varying shapes and dimensions of buildings are among other challenges that hinder present day models in segmenting sharp building boundaries. High-quality aerial imagery datasets facilitate comparisons of existing methods and lead to increased interest in aerial imagery applications in the machine learning and computer vision communities.

Content

The Massachusetts Buildings Dataset consists of 151 aerial images of the Boston area, with each of the images being 1500 × 1500 pixels for an area of 2.25 square kilometers. Hence, the entire dataset covers roughly 340 square kilometers. The data is split into a training set of 137 images, a test set of 10 images and a validation set of 4 images. The target maps were obtained by rasterizing building footprints obtained from the OpenStreetMap project. The data was restricted to regions with an average omission noise level of roughly 5% or less. The large amount of high quality building footprint data was possible to collect because the City of Boston contributed building footprints for the entire city to the OpenStreetMap project. The dataset covers mostly urban and suburban areas and buildings of all sizes, including individual houses and garages, are included in the labels. The datasets make use of imagery released by the state of Massachusetts. All imagery is rescaled to a resolution of 1 pixel per square meter. The target maps for the dataset were generated using data from the OpenStreetMap project. Target maps for the test and validation portions of the dataset were hand-corrected to make the evaluations more accurate.

Refer this thesis for more information.

Acknowledgements

This dataset is derived from Volodymyr Mnih's original Massachusetts Buildings Dataset. Massachusetts Roads Dataset & Massachusetts Buildings dataset were introduced in Chapter 6 of his PhD thesis. If you use this dataset for research purposes you should use the following citation in any resulting publications:

@phdthesis{MnihThesis, author = {Volodymyr Mnih}, title = {Machine Learning for Aerial Image Labeling}, school = {University of Toronto}, year = {2013} }

Inspiration

Rapid advances in Image Understanding using Computer Vision techniques have brought us many state-of-the-art deep learning models across various benchmark datasets. Can we better address the challenges faced by the current models in segmenting buildings from aerial images using the latest methods? Do state-of-the-art methods from other benchmarks work equally well on this data? Does engineering features specific to buildings datasets allow us to build better models?

Go to Massachusetts Roads Dataset.

Low-altitude (30-120 meters above ground level) digital images of Town Neck Beach in Sandwich, Massachusetts, were obtained with a series of cameras mounted on small unmanned aerial systems (UAS, also known as a drone). Imagery was collected at close to low tide on five days to observe changes in beach and dune morphology. The images were geolocated by using the single-frequency geographic positioning system aboard the UAS. Ground control points (GCPs) were established by using temporary targets on the ground, which were located by using a real-time kinematic global navigation satellite system (RTK-GNSS) base station and rovers. The GCPs can be used as constraints during photogrammetric processing. Transect points were collected by using the same RTK-GNSS system; these can be used to evaluate photogrammetric products. This data release includes georeferenced images, image-location files, GCPs, and transect points. Collection of these data was supported by the U.S. Geological Survey Coastal/Marine Hazards and Resources Program and conducted under U.S. Geological Survey field activity numbers 2016-013-FA, 2016-043-FA, 2016-053-FA, 2016-054-FA and 2016-055-FA.

This dataset is part of the larger data collection, “Aerial imagery object identification dataset for building and road detection, and building height estimation”, linked to in the references below and can be accessed here: https://dx.doi.org/10.6084/m9.figshare.c.3290519. For a full description of the data, please see the metadata: https://dx.doi.org/10.6084/m9.figshare.3504413.

Imagery data from the United States Geological Survey (USGS); building and road shapefiles are from OpenStreetMaps (OSM) (these OSM data are made available under the Open Database License: http://opendatacommons.org/licenses/odbl/1.0/); and the Lidar data are from U.S. National Oceanic and Atmospheric Administration (NOAA), the Texas Natural Resources Information System (TNRIS).

The National Aerial Photography Program (NAPP) was coordinated by the USGS as an interagency project to acquire cloud-free aerial photographs at an altitude of 20,000 feet above mean terrain elevation. The photographs were taken with a 6-inch focal length lens at a scale of 1:40,000. Coverage over the conterminous United States includes both black-and-white (BW) and color infrared (CIR) aerial photographs. Film type and extent of coverage were determined by available funds and operational requirements. The NAPP program, which was operational from 1987 to 2007, consists of more than 1.3 million images. Photographs were acquired on 9-inch film and were centered over quarters of USGS 7.5-minute quadrangles.To view historical imagery availability by county please visit the Historical Availability of Imagery map.To view more NAPP imagery visit the NAPP Historical Imagery Portfolio app.For ordering information please contact the GEO Customer Service Section at geo.sales@usda.gov.

Imagery acquired with unmanned aerial systems (UAS) and coupled with structure from motion (SfM) photogrammetry can produce high-resolution topographic and visual reflectance datasets that rival or exceed lidar and orthoimagery. These new techniques are particularly useful for data collection of coastal systems, which requires high temporal and spatial resolution datasets. The U.S. Geological Survey worked in collaboration with members of the Marine Biological Laboratory and Woods Hole Analytics at Black Beach, in Falmouth, Massachusetts to explore scientific research demands on UAS technology for topographic and habitat mapping applications. This project explored the application of consumer-grade UAS platforms as a cost-effective alternative to lidar and aerial/satellite imagery to support coastal studies requiring high-resolution elevation or remote sensing data. A small UAS was used to capture low-altitude photographs and GPS devices were used to survey reference points. These data were processed in an SfM workflow to create an elevation point cloud, an orthomosaic image, and a digital elevation model.

The U.S. Geological Survey (USGS) conducts baseline and storm response photography missions to document and understand the changes in vulnerability of the Nation's coasts to extreme storms. On May 19-22, 2009, the USGS conducted an oblique aerial photographic survey from Owls Head, Maine, to the Virginia/North Carolina border, aboard a Cessna 207A aircraft at an altitude of 500 feet (ft) and approximately 1,200 ft offshore (Figure 2, http://pubs.usgs.gov/ds/0946/html/ds946_fig2.html). This mission was flown to collect data for assessing incremental changes since the last survey, flown October 2000, and can be used for assessing future coastal change. The photographs provided here are Joint Photographic Experts Group (JPEG) images. The photograph locations are an estimate of the position of the aircraft and do not indicate the location of the feature in the images. (See the Navigation Data page in the corresponding report, for additional details, http://pubs.usgs.gov/ds/0946/html/ds946_nav.html). These photographs document the configuration of the barrier islands and other coastal features at the time of the survey. ExifTool (http://www.sno.phy.queensu.ca/~phil/exiftool/) is a free software program for reading, writing, and manipulating image, audio, and video metadata. ExifTool was used to add the following to the header of each photo: time of collection, Global Positioning System (GPS) latitude, GPS longitude, keywords, credit, artist (photographer), caption, copyright, and contact information. Photographs can be opened directly with any JPEG-compatible image viewer by clicking on a thumbnail on the contact sheet. All image times are recorded in UTC. Table 1 (http://pubs.usgs.gov/ds/0946/html/ds946_table.html) provides detailed information about the assigned location, name, date, and time the photograph was taken along with links to the photograph. In addition to the photographs, a Google Earth Keyhole Markup Language (KML) file is provided and can be used to view the images by clicking on the marker and then clicking on either the thumbnail or the link above the thumbnail. The KML files were created using the photographic navigation files. Note: A KML number was assigned to each photograph to aid navigation of the Google Earth file. These numbers correspond to the site labels in Google Earth.

These medium resolution true color (RGB - Red-Green-Blue bands) images represent the first color aerial photo "basemap" for the Commonwealth from MassGIS. MassGIS (then part of the Executive Office of Environmental Affairs) and the Massachusetts Dept. of Transportation (then the Mass Highway Department) jointly funded the project. The photography for the mainland was captured in April 2001 when deciduous trees were mostly bare and the ground was generally free of snow. Photography for the Elizabeth Islands, Martha's Vineyard, and Nantucket (the "Islands") was captured in April 2003. Imagery is available for the entire state. Original imagery pixel resolution is 1/2-meter.For full metadata visit https://www.mass.gov/info-details/massgis-data-2001-2003-aerial-imagery.

Spring 2019

Imagery acquired with unmanned aerial systems (UAS) and coupled with structure from motion (SfM) photogrammetry can produce high-resolution topographic and visual reflectance datasets that rival or exceed lidar and orthoimagery. These new techniques are particularly useful for data collection of coastal systems, which requires high temporal and spatial resolution datasets. The U.S. Geological Survey worked in collaboration with members of the Marine Biological Laboratory and Woods Hole Analytics at Black Beach, in Falmouth, Massachusetts to explore scientific research demands on UAS technology for topographic and habitat mapping applications. This project explored the application of consumer-grade UAS platforms as a cost-effective alternative to lidar and aerial/satellite imagery to support coastal studies requiring high-resolution elevation or remote sensing data. A small UAS was used to capture low-altitude photographs and GPS devices were used to survey reference points. These data were processed in an SfM workflow to create an elevation point cloud, an orthomosaic image, and a digital elevation model.

Basemap from MassGIS using the Massachusetts 2023 Color Aerial Imagery tile layer. This basemap appears in the MassGIS Basemap Gallery in ArcGIS Online and ArcGIS Pro.

Aerial photographs were taken of the aquatic vegetation of Florida Bay, Indian River (Florida), and the Coast of Massachusetts. Photographs were scanned and geo-referenced for the purpose of mapping. Data is contained on a "DLT" tape and is stored "off-site" as a secure backup copy.

This dataset contains images obtained from unmanned aerial systems (UAS) flown in the Cape Cod National Seashore. The objective of the field work was to evaluate the quality and cost of mapping from UAS images. Low-altitude (approximately 120 meters above ground level) digital images were obtained from cameras in a fixed-wing unmanned aerial vehicle (UAV) flown from the lawn adjacent to the Coast Guard Beach parking lot on 1 March, 2016. The UAV was a Skywalker X8 flying wing operated by Raptor Maps, Inc., contractors to the U.S. Geological Survey. U.S. Geological Survey technicians deployed and mapped 28 targets that appear in some of the images for use as ground control points. All activities were conducted according to Federal Aviation Administration regulations and under a National Park Service Scientific Research and Collecting Permit, study number CACO-00285, permit number CACO-2016-SCI-003. Two consecutive UAS missions were flown, each with two cameras, autopilot computer, radios, and a global navigation satellite system (GNSS) positioning system as payload. The first flight (f1) was launched at approximately 1112 EST, and followed north-south flight lines, landing at about 1226 EST. Two Canon Powershot SX280 12-mexapixel digital cameras, designated rgb1 and rgb2 made images during this flight. The second flight (f2) was launched at 1320 EST and followed east-west flight lines, landing at 1450 EST. Prior to f2, rgb2 was replaced with a Canon SX280 modified with a Schott BG 3 filter to emphasize light at near-infrared wavelengths, designated nir1. Rgb1 and nir1 made images during this second flight. In addition to the images, this dataset also contains locations of both in-situ and placed targets that may be used as ground control to constrain photogrammetric reconstructions.

Small Uncrewed Aircraft Systems (sUAS) were used to collect aerial remote sensing data over Sesuit Marsh in Dennis, MA. Raw data from aerial surveys include aerial images from natural color (RGB) and multispectral cameras and raw lidar data. These datasets were processed to produce high resolution digital elevation models (DEM), image mosaics, and lidar point clouds (LPC) to provide local partners with high resolution imagery and elevation data to monitor and identify vegetation cover of a salt marsh on Cape Cod during peak growing season. A YellowScan Mapper Plus lidar system with integrated RGB camera module were mounted to a DJI Matrice 600 to collect the raw lidar data and true-color imagery. A MicaSense Altum-PT was then mounted to the DJI Matrice 600 to collect the multispectral reflectance imagery. Black and white cross-coded ground control points (GCPs) were surveyed using RTK-GPS and RTK-enabled AeroPoints to georeference the model and orthomosaics during post-proces ...

This reference contains the imagery data used in the completion of the baseline vegetation inventory project for the NPS park unit. Orthophotos, raw imagery, and scanned aerial photos are common files held here. High-quality existing photography developed by MassGIS was used as the base for the vegetation map (MassGIS 2007). A true color orthophoto mosaic was developed from a set of digital 1:5,000 scale medium resolution true color aerial images that are considered the new "base map" for the Commonwealth of Massachusetts by MassGIS and the Executive Office of Environmental Affairs (EOEA). The photography for the entire commonwealth was captured in April 2005 when deciduous trees were mostly bare and the ground was generally free of snow. The image type is 4-band (RGBN) natural color (Red, Green, Blue) and Near infrared in 8 bits (values ranging 0–255) per band format. Key information for the mosaic is summarized in Table 1. Appendix A contains additional detailed information regarding the aerial photography acquired from MASS GIS.

In spring 2008, the U.S. Geological Survey, as part of its Boston 133 Cities Urban Area mapping program, contracted for true-color imagery covering the metropolitan Boston area and beyond. Image type for the entire region (more than 1.7 million acres) is 24-bit, 3-band (red, green, blue) natural color. Each band has pixel values ranging 0-255. Pixel resolution is 30 cm., or approximately one foot. In spring 2009, USGS continued the project and 4-band 30cm imagery was obtained for the remainder of the state.This digital orthoimagery can serve a variety of purposes, from general planning, to field reference for spatial analysis, to a tool for data development and revision of vector maps. It can also serve as a reference layer or basemap for myriad applications inside geographic information system (GIS) software.The data are served from MassGIS' ArcGIS Online account as a tiled cached map service for fast display.For full metadata and links to download the imagery visit https://www.mass.gov/info-details/massgis-data-20082009-aerial-imagery.